90 AN ANALYSIS OF TESTS OF WATER-TIGHT BULKHEADS. 



denced by the great permanent set, ij inches in 9 feet, this case is rather 

 exceptional. (See table of stresses and deflections, page 94.) 



The next question to be examined is whether brackets, when such are 

 fitted, shall be considered effective, and to what degree, or in other words, 

 which formula will in each case most nearly conform to the actual condi- 

 tions, the formula for a stiffener fixed at the ends or that for a stiffener 

 freely supported at the ends. 



In order to settle this point, curves of deflection have been calculated 

 according to both of these formulas, and these curves are shown in light 

 lines on the diagrams. A comparison between the observed and the calcu- 

 lated curves will at once show which of the two assumptions is nearest 

 the truth, but due regard must be paid to the rigidity of the structure 

 to which the bracket is attached. 



The length, /, of the stiffeners chosen for the calculations of this analy- 

 sis was, in case of the formula for both ends fixed, the actual length of the 

 main bar of the stiflener, denoted by /, and the same length was used 

 where one end was unbracketed and the other efficiently bracketed, in 

 which case the formula for one end free, the other end fixed, was employed. 

 In the formula for both ends freely supported, when employed to stiffeners 

 bracketed at both ends, the length, I2, was measured between the apices 

 of the brackets. This is clearly indicated on the diagrams. This choice 

 of length is of course arbitrary, but has been found to lead to a fair corre- 

 spondence with the observed results in case of double stiffeners. In com- 

 paring the curves it is well to bear in mind that the deflection varies as 

 the fourth or fifth power of tlie length. 



In calculating the moment of inertia of the double stiffeners, the bulk- 

 head plating, which is here placed at the neutral axis, has been neglected, 

 but in case of the single stiffeners it was found that a certain strip of the 

 bulkhead plating ought to be considered as forming part of and as working 

 with the stiffeners. If this strip was not included, the curve of deflection, 

 calculated for both ends fixed, would, in case of the single stiffeners of the 

 United States ships, fall considerably outside the curve for the observed 

 elastic deflection even in the largest bulkheads. This does not seem pos- 

 sible, for since absolute rigidity in the brackets and in the supporting 

 structure can never be secured, we should rather expect the calculated 

 curve to fall inside the observed curve. It appears, therefore, that the 

 moment of inertia of the stift'eners has been underestimated, but by includ- 

 ing a strip of bulkliead plating of a certain width, the moment of inertia 

 may be increased so as to make the calculated curve of deflection coincide 

 with the observed curve. The width of this strip in the present analysis 



